PSI - Issue 44

Andrea Ciancimino et al. / Procedia Structural Integrity 44 (2023) 323–330 Andrea Ciancimino et al. / Structural Integrity Procedia 00 (2022) 000–000

329

7

A progressive accumulation of permanent crest settlement w is observed as a result of progressive soil densification and earthquake-induced temporary sliding instabilities along the slopes of the dams (Fig.4c-d). Such an accumulation rapidly increases in correspondence with the pore water pressure build-up, while it tends to stabilize afterwards when the global equilibrium of the structures is re-established. Fig. 5 reports the comparison between simulated w s and predicted w p settlement for the two dams. The predictions are computed according to the widely-used Swaisgood’s (2003) equation for estimating the crest settlement normalized to the height of the dam H d plus the foundation thickness H f :

H

D

( ) %

d

w

e

=

=

(2)

PGA M +

6.07

0.57 8 × -

×

p

d H H +

f

where PGA is the peak ground acceleration of the input motion in g units, while M is the corresponding magnitude. The empirical predictions are, on average, in good accordance with the numerical results for the Arignano dam, although the relationship predicts a smaller variation of the expected settlement with the input intensity measure. On the other side, the obtained w s for the Briaglia dam ranging from 0.1% to 10% are systematically underestimated by Swaisgood’s (2003) equation, which instead predicted an almost constant w p equal to about 0.01%. In an attempt of improving the predictions given by the simplified approach, a further empirical relationship is calibrated through a nonlinear least-squares procedure based on the results of the numerical simulations. The proposed equation is inspired by the original Swaisgood’s (2003) equation and can be expressed in the following general form: being A and a calibration parameters. The parameter a defines the slope of the relationship on a logarithmic scale and is thus associated with the dependency of the predicted settlement from the chosen seismic intensity measure. In this study, the PGA is selected for the sake of simplicity. The adopted value of A seems instead to be related to the specific features of the dam, in terms of geometric characteristics and different thicknesses of the soil layers prone to pore water pressure build-up. As a consequence, the same value of the parameter a , equal to 27.1, is adopted for the two dams. Conversely, an A parameter equal to 5·10 -4 and 0.11, is employed respectively for the Arignano and Briaglia dams. The comparison between w s and w p computed through Eq. 3 is again presented in Fig. 5. The relationship seems able to properly predict both the variation of the expected crest settlement with increasing seismicity levels and the specific features of the two dams. ( ) % PGA p w A e a× = × (3)

Fig. 5. Simulated versus predicted settlement computed according to Swaisgood’s (2003) equation and Eq. 3.